Study characteristics

Figure S1. Data coverage by geography

Number of study-periods in the primary dataset at each administrative level by country. An individual study has >1 “study-periods” if it reports cholera positivity for >1 time frame. Countries with >10 study-periods displayed as 10.


Figure S2. Data coverage over time

Number of study-location-periods in the primary dataset at each administrative level by country within different time periods. An individual study has >1 “study-location-periods” if it reports cholera positivity for >1 time frame and/or for >1 administrative level. Countries with >10 study-location-periods displayed as 10. Year represents the year sampling was completed. Excludes 8 studies missing a study end date.


Table 1. Study characteristics

Number of study-observations included in the dataset with each study characteristic. An individual study has >1 study-observations if it reports V. cholerae positivity for >1 sampling method or country.

variable value Total Percent
Study design Diagnostic test accuracy 19 15.2
Study design Other 1 0.8
Study design Surveillance 96 76.8
Study design Vaccine effectiveness 9 7.2
Sampling quality High 37 29.6
Sampling quality Low 88 70.4
Prop lab tested 0-5 15 12.0
Prop lab tested 5-50 31 24.8
Prop lab tested 50-95 22 17.6
Prop lab tested 95+ 57 45.6
Num tests 1 98 78.4
Num tests 2 20 16.0
Num tests 3+ 7 5.6
N 1-9 1 0.8
N 10-99 37 29.6
N 100-999 62 49.6
N 1000+ 25 20.0


Table S2. Characteristics of suspected cholera cases reported by each study

variable value Total Percent
Prop vaccinated No 115 95.8
Prop vaccinated Yes 5 4.2
Prop under 5 No 86 71.7
Prop under 5 Yes 34 28.3
Prop under 15 No 99 82.5
Prop under 15 Yes 21 17.5
Prop severe dehyd No 110 91.7
Prop severe dehyd Yes 10 8.3
Prop mod dehyd No 111 92.5
Prop mod dehyd Yes 9 7.5
Prop female No 93 77.5
Prop female Yes 27 22.5
Prop antibiotics No 110 91.7
Prop antibiotics Yes 10 8.3
Prop 5 and severe dehyd No 115 95.8
Prop 5 and severe dehyd Yes 5 4.2
Prop 5 and dehyd and antibiotics No 119 99.2
Prop 5 and dehyd and antibiotics Yes 1 0.8


V. cholerae positivity in unadjusted data

Figure 2. Vibrio cholerae positivity by study methodology and outbreak context

Proportion of suspected cholera cases that were confirmed positive by A) diagnostic test type, B) quality of sampling methods, where “high” includes all suspected cases or a random or stratified sample and “low” includes convenience or unreported sampling methods, C) age minimum in suspected case definition, where “0” indicates that no minimum age was set, and D) whether surveillance was initiated in response to an outbreak or whether it was routine surveillance or non-outbreak. Each point is a study-observation; individual studies may have multiple points if they reported positivity by multiple tests, reported results for multiple countries or outbreak contexts, and/or multiple sampling methods. Circles represent study-observations with high quality sampling and triangles indicate low quality sampling.


Figure S3. Vibrio cholerae positivity by incidence and suspected case characteristics

Relationship between reported V. cholerae positivity and A) proportion of suspected cholera cases tests that were under 5 years of age, B) suspected cholera incidence rate per 10,000 at study site, C) proportion of suspected cases severely dehydrated, and D) proportion of suspected cases on antibiotics prior to testing. Size of the points is proportional to the number of cases tested, and shapes indicate which diagnostic test was used to confirm V. cholerae infection. Confidence intervals for Spearman rank correlation coefficients estimated using bootstrapping (nrep=1000). Smoothing method is loess (without weights). In B) estimated suspected cholera incidence rates for 2010-2016 (Lessler et al., 2018) were aggregated to the administrative division that best represented each study’s catchment area by dividing the total estimated cholera cases in each area by its estimated population.


Adjusted underlying V. cholerae positivity

Table S3. Estimated sensitivity and specificity of each diagnostic test

Estimate is median sensitivity and specificity pooled across four studies that reported results for all diagnostics tests, as described in Methods. Parentheses show 95% Credible Interval.

Measure Test Estimate (%)
Sensitivity Culture 82 ( 37.5 - 98.7 )
Sensitivity PCR 85.1 ( 53.6 - 98.9 )
Sensitivity RDT 90.4 ( 55.2 - 99.5 )
Specificity Culture 94.3 ( 81.5 - 99.6 )
Specificity PCR 94.2 ( 81.1 - 99.7 )
Specificity RDT 88.9 ( 54.8 - 99.4 )


Figure S4. Posterior distributions of V. cholerae positivity

Posterior distributions of underlying true V. cholerae positivity estimated using the random-effects model adjusted for diagnostic test sensitivity and specificity and study methods (i.e., sampling quality and age minimum in case definition), corresponding to results in Table 2. A) Prior (green) and posterior distribution (purple) of positivity in logit space. B) Prior (green) and posterior predictive distribution (purple) of positivity in logit space. C) Histogram of estimated positivity, or the true underlying proportion positive, based on studies in the meta-analysis. D) Histogram of the posterior predictive distribution of positivity, or what we predict the underlying proportion positive would be in a new hypothetical study site.


Table S4. Estimated underlying V. cholerae O1/O139 positivity

“Unadjusted” is mean percent positive (95% confidence interval) from random effects meta-analysis implemented using R package “meta” without any adjustments. “Adjusted” is the underlying true percent positive (95% credible interval) estimated using the Bayesian hierarchical model, adjusted for sensitivity/specificity of the tests, age in case definition, sampling quality, and whether or not surveillance was initiated in response to an outbreak. Also includes sensitivity analysis of increasing variance on all priors

Version Positivity (%)
Unadjusted 41 ( 37 - 46 )
Adjusted for test performance 48 ( 42 - 54 )
Adjusted for test performance & methods 49 ( 43 - 55 )
Adjusted for test performance, methods, & outbreak setting 49 ( 43 - 56 )
Adjusted for test performance, methods, & outbreak setting, increased variance on priors 49 ( 43 - 56 )


Figure 3. Estimated underlying V. cholerae positivity

A) Posterior distributions of pooled percent sensitivity and specificity of culture (top), PCR (middle), and RDT (bottom) for detecting V. cholerae O1O139 infections in suspected cholera cases. Dashed lines represent median values of each distribution. B) “Unadjusted” is mean percent positive (95% confidence interval) from random effects meta-analysis implemented using R package “meta” without any adjustments. “Adjusted” is the underlying true percent positive (95% credible interval) estimated using the Bayesian hierarchical model, adjusted for sensitivity/specificity of the tests, age in case definition, sampling quality, and whether or not surveillance was initiated in response to an outbreak.


Figure 4. Forest plot of study estimates and underlying positivity

Black points indicate median study-level underlying positivity and 95% Credible Interval (CI). Teal, orange, and purple points indicate the proportion positive reported by study for culture, PCR, and RDT, respectively, and corresponding error bars indicate 95% confidence interval for a binomial probability. Studies are labeled by country ISO3 code, whether or not they used high quality sampling methods, and whether a minimum age was set in the suspected cholera case definition. Green distribution indicates median and 95% CI of pooled positivity. Pink distribution indicates median and 95% CI of posterior predictive distribution of positivity. Studies are split into outbreak and non-outbreak for ease of interpretation, but the green and pink distributions represent the global estimates across all studies.

## I2: 99.99; tau2: 0.79


Factors associated with variation in V. cholerae positivity

Table S5. Odds of V. cholerae positivity by age and outbreak context

Odds that a suspected cholera case seeking testing or care has a true V. cholerae O1/O139 infection with low sampling quality compared to high sampling quality, with increasing minimum age set in suspected case definition, and with surveillance initiated in response to an outbreak compared to non-outbreak surveillance (i.e., routine or post-vaccination surveillance).

Variable Odds ratio
Low sampling quality 1.48 ( 0.92 - 2.38 )
Minimum age in case definition 1.16 ( 1.01 - 1.34 )
Outbreak surveillance 1.64 ( 1.06 - 2.52 )


Table S6. Odds of V. cholerae positivity by categorical age in case definition

Odds that a suspected cholera case seeking care has a true V. cholerae infection by minimum age set in suspected case defintion. Coefficients correspond to random effects models run with age as a categorical variable (reference group = no age minimum age explicitly set in case definition or 0). All other model settings and covariates are the same as in Table 3.

Variable Odds ratio
Low sampling quality 1.47 ( 0.91 - 2.47 )
Min age in case definition: 1 1.19 ( 0.48 - 2.98 )
Min age in case definition: 2 1.23 ( 0.66 - 2.28 )
Min age in case definition: 5 1.46 ( 0.64 - 3.19 )
Min age in case definition: 10 2.57 ( 0.52 - 12.75 )
Outbreak surveillance 1.7 ( 1.09 - 2.65 )


Table S7. Sensitivity analysis of the odds of V. cholerae positivity by age and outbreak context

Odds that a suspected cholera case seeking testing or care has a true V. cholerae O1/O139 infection with low sampling quality compared to high sampling quality, with increasing minimum age set in suspected case definition, and with surveillance initiated in response to an outbreak compared to non-outbreak surveillance (i.e., routine or post-vaccination surveillance) in models where variance was increased on all priors.

Variable Odds ratio
Low sampling quality 1.48 ( 0.92 - 2.38 )
Minimum age in case definition 1.16 ( 1.01 - 1.34 )
Outbreak surveillance 1.64 ( 1.06 - 2.52 )